LLM & Agents

Enable AI agents to introspect and control Castella UIs via MCP. Create MCP servers, expose UI resources, handle MCP tools, and use semantic IDs.

Deduplicate and consolidate CLAUDE.md memory files. Trigger when user says "condense my CLAUDE.md files", "deduplicate CLAUDE.md", "clean up my memory files", or "consolidate my instructions". Removes redundancy within files and across the file hierarchy.

Add or update automated tests for a change (bugfix, feature, refactor) using the repository’s testing rules in `AGENTS.md`. Use TDD where applicable; derive scenarios from docs/Features/* and ADR invariants; prefer stable integration/API/UI tests, run build before tests, and verify meaningful assertions for happy/negative/edge cases.

Use to inspect qualification coverage across Metrics, Economic buyer, Decision criteria, Decision process, Implications, Champion.

Dashboard pattern for tracking doc coverage across product versions, locales, and channels.

Get Gemini's code review of git changes after Claude makes edits. Trigger when user wants a second opinion on code changes ("have Gemini review my changes", "get code review from Gemini", "review this diff"), or as a final check before committing.

Get Gemini's review of Claude's implementation plans. Trigger when user wants a second opinion on a plan ("have Gemini review this plan", "get second opinion", "critique this plan"), or after Claude creates a plan file that needs validation before implementation.

Use to run day-to-day loyalty reward catalog management and fulfillment.

Comprehensive Supabase expert with access to 2,616 official documentation files covering PostgreSQL database, authentication, real-time subscriptions, storage, edge functions, vector embeddings, and all platform features. Invoke when user mentions Supabase, PostgreSQL, database, auth, real-time, storage, edge functions, backend-as-a-service, or pgvector.

Use to enforce cadence rules, timezone coverage, and operational controls on social calendars.

Connect to A2A protocol agents from Castella. Communicate with agents, display agent cards, send messages, handle responses, and stream results.

Format code and keep style consistent using the repository’s canonical formatting/lint commands from `AGENTS.md`. Use after implementing changes or when formatting drift causes noisy diffs; keep formatting changes intentional and verified with build/tests.

Editing mode for clarity and semantic density; use for prompts, docs, specs, emails, and naming.

使用 git worktree + 背景代理会话管理大型任务（默认启动 Claude Code；可用 --codex 快速切到 Codex CLI，或通过 --agent-cmd 指定任意命令）。适合用户希望在不阻塞当前会话的情况下执行大型或多步骤任务。

Orchestrate multi-agent swarms with agentic-flow (roles, topology, modes, shared state, resiliency).

Create and manage Claude Code skills following Anthropic best practices. Use when creating new skills, modifying skill-rules.json, understanding trigger patterns, working with hooks, debugging skill activation, or implementing progressive disclosure. Covers skill structure, YAML frontmatter, keyword triggers, hook mechanisms (UserPromptSubmit), session tracking, and the 500-line rule.

Prioritizes tests based on ROI and critical user journeys. Use when: deciding test coverage, selecting test types, or budgeting effort.

Testing commands and patterns for Oak AI project including running Jest unit tests, E2E tests with Playwright, coverage reports, and filtering tests. Use when writing tests, running test suites, or debugging test failures.

Retrieval-augmented generation (RAG) skill for the D&D 5e System Reference Document (SRD). Use when answering questions about D&D 5e core rules, spells, combat, equipment, conditions, monsters, and other SRD content. This skill provides agentic search-based access to the SRD split into page-range markdown files.

Creates summary for human review - and comprehensive implementation plan for LLM, complete code examples, and verification steps